a b s t r a c t

Background: It is suggested that a prehospital scale should be utilized to identify patients with emergent Large vessel occlusion (ELVO). We aimed to perform external validation of nine ELVO scales.

Methods: This single center retrospective observational study included patients with ischemic stroke visiting the emergency department (ED) within 6 h of symptom onset. Participants were excluded if individual items of the National Institute of Health Stroke Scale scores were not recorded or they did not receive brain computed tomog- raphy angiography or magnetic resonance imaging before intravenous thrombolysis or endovascular thrombectomy, and within 24 h of ED admission. The first definition of ELVO was emergent occlusion of the in- ternal carotid artery (ICA) and Middle cerebral artery segment 1 (M1). The second definition was emergent oc- clusion of ICA, M1, Basilar artery, middle cerebral artery segment 2, anterior cerebral artery segment 1, and posterior cerebral artery segment 1. Area under the Receiver operating characteristic curve was con- structed to examine discrimination. The sensitivity, specificity, positive predictive value, and negative predictive value of the nine scales under the two ELVO definitions were calculated.

Results: A total of 1231 patients were included in the study. No significant differences were observed in the

AUROC under the two ELVO definitions. However, sensitivity values of these scales were largely different, ranging from 44.56% to 93.68% under the first ELVO definition. The sensitivity values among scales were also different under the second ELVO definition.

Conclusion: Stakeholders in the community should choose suitable scales according to their own system conditions.

(C) 2020

1. Introduction

Endovascular thrombectomy (EVT) has been shown to be beneficial for patients with emergent large vessel occlusion (ELVO) [1-5]. It has also been shown that the shorter the time interval between the onset of symptoms and reperfusion using EVT, the better the functional out- come at three months [6-9]. Interhospital transfer delays the time to re- ceive EVT. It makes the patients less likely to have independent ambulation at discharge or to be discharged home and a higher rate of symptomatic intracranial hemorrhage [10,11]. Both the Standards and Guidelines Committee of the Society of Neurointerventional Surgery

* Corresponding author at: Department of Emergency Medicine, National Taiwan University Hospital, No. 7 Chung Shan South Rd., Taipei 100, Taiwan.

E-mail address: [email protected] (M.-J. Hsieh).

and the guidelines of the American Heart Association/American Stroke Association (AHA/ASA) recommend that effective Stroke severity scale should be established to identify patients with a strong probability of ELVO stroke who are ineligible for intravenous thrombolysis. This is to facilitate the rapid transport of patients potentially eligible for thrombectomy to the closest facilities with EVT [12,13]. Therefore, in this age of EVT usage, it is important to determine an effective tool to identify patients with ELVO in a prehospital setting.

Many prehospital stroke scales developed for ELVO recognition have been reported in literature [14-18]. These scales have a high accuracy and can also reach a balance between sensitivity and specificity [19]. In- correct assessment of the patients with ELVO and sending them to a fa- cility without the capability of EVT may lead to delay or even miss the golden reperfusion time, while misdiagnosing a minor stroke patient as ELVO could lead to the overload of a facility with EVT or even delay

https://doi.org/10.1016/j.ajem.2020.12.011

0735-6757/(C) 2020

the time for intravenous thrombolysis administration. Prehospital scales need to be easy enough for providers in the emergency medical service (EMS) system to remember and quickly perform in a prehospital setting.

Although many prehospital scales for identifying patients with ELVO have been developed and embedded in some systems elsewhere [20,21], there is currently no external validated scale in Taiwan. Plans are being put in place in the EMS system of Taiwan to include a bypass strategy into the protocol for patients with a suspected diagnosis of ELVO. Therefore, our study aimed to perform external validation for nine ELVO scales using clinical data of patients in Taiwan to provide sci- entific evidence in choosing a scale that best fits our EMS system.

1. Methods
   1. Source of data

In Taiwan, the average annual incidence rate of stroke is about 330 per 100,000 population [22]. The proportion of ischemic stroke (IS) is 74% of all stroke cases [23]. This single center observational study retro- spectively analyzed a prospective Stroke registry at National Taiwan University Hospital (NTUH), a Tertiary medical center with 2450 beds in Taipei, Taiwan. In this hospital, we started to investigate the risk fac- tors, clinical course, prognosis, and complications in different types of acute strokes (such as IS), intracerebral hemorrhage (ICH), subarach- noid hemorrhage (SAH), Transient ischemic attack , and cerebral venous thrombosis in 1995. Patients who had stroke onset within 10 days of hospital admission or during hospitalization were enrolled in the registry [24,25]. This study and stroke registry were approved by the Institutional Review Board of NTUH.

Data from the NTUH stroke registry were collected by three full-time study nurses. After a patient was diagnosed with stroke or TIA according to the definition of the World Health Organization (WHO) [26], the pa- tient was invited to join the registry with a written informed consent. If the patient was unconscious, the next of kin provided the consent. Med- ical and other associated information of the patient was gathered pro- spectively with a recording form through a direct query with the patient and relatives, as well as a review of medical record. The record- ing form of the NTUH stroke registry adopted that of the Taiwan stroke registry since 2006, which was well delineated in a previous study [27]. The National Institutes of Health Stroke Scale score, including the scores of each item and the total scores were evaluated by a resident neurologist or a board-certified neurologist and collected by the full- time study nurses. In addition, data of brain imaging including com- puted tomography angiography (CTA) and magnetic resonance imaging (MRI), which can be used to evaluate intracranial and carotid vessel conditions, were also included. Each brain image was evaluated by a board-certified neuroradiologist who provided the final report.

• 1. Inclusion and exclusion criteria

In this study, we analyzed the data of patients with IS from the reg- istry who visited the emergency department (ED) within 6 h of symp- tom onset between January 1, 2013, and September 30, 2018. The exclusion criteria were as follows: (1) patients with incomplete record or no evaluation of individual item of NIHSS scores; (2) patients who did not undergo CTA or MRI of the brain before intravenous thromboly- sis or EVT; and (3) patients who did not receive intravenous thrombol- ysis or EVT and CTA or MRI of the brain within 24 h of ED admission.

• 1. Definition of outcome

The outcome of interest was the presence of ELVO in brain images. The reviewers were blinded to follow-up clinical findings but had infor- mation on the age and sex of the patients. We adopted two kinds of ELVO definitions in our study. The first definition was acute occlusion

of the Internal carotid artery (ICA) and the middle cerebral artery seg- ment 1 (M1). The second definition was acute occlusion of the ICA, M1, basilar artery (BA), middle cerebral artery segment 2 (M2), anterior cerebral artery segment 1 (A1), and posterior cerebral artery segment 1 (P1). Although the AHA/ASA guidelines recommends EVT within 6 h of stroke onset for patients with causative occlusion of ICA and M1, guide- lines also suggested EVT for patients with BA, M2, A1, and P1 [13]. In some hospitals in Taiwan including our hospital, patients with causative occlusion of BA, M2, A1, and P1 are taken into consideration to benefit from EVT and would receive EVT if other criteria were met [28]. There- fore, it is important to understand the capability of the prehospital scales used in identifying these patients.

• 1. Predictors

In this study, we compared the capability of nine different prehospital stroke scales for predicting ELVO: 3-Item Stroke Scale (3-ISS) [15], Emergency Medical Stroke Assessment (EMSA) [29], Cin- cinnati Prehospital stroke severity scale (CPSSS) [17], Vision, Aphasia, Neglect (VAN) screening tool [18], Field Assessment Stroke Triage for Emergency Destination (FAST-ED) [30], Prehospital Acute Stroke Sever- ity Score (PASS) [31], Rapid Arterial occlusion Evaluation (RACE) [32], Finnish Prehospital Stroke Scale (FPSS) [33], and Gaze-Face-Arm- Speech-Time test (G-FAST) [34]. We used the cut-off value of each scale as suggested in the original studies. These prehospital stroke scales were initially developed from the individual items of NIHSS. Therefore, we adopted the corresponding scores of individual NIHSS items of every patient to match the items of these prehospital stroke scales. The de- tailed NIHSS correspondence to the items of the prehospital stroke scales is shown in Supplementary Table S1. The total points and cutoff values of the scales are shown in Supplementary Table S2.

• 1. Sample size

In one study, ELVO was detected in 23.6% of all patients with IS along with large vessel definition of ICA and M1 [34]. We assumed that ELVO was detected in 23% of all patients with IS in our study cohort. In another study, it was suggested that the external validation of a prognostic model ideally requires over 200 events [35]. Therefore, at least 870 pa- tients with IS were needed for our study.

• 1. Statistical analysis

The area under the receiver operating characteristic curve (AUROC) and the respective 95% confidence intervals (CIs) were used to assess the performance of different prehospital stroke scales to predict ELVO. The sensitivity, specificity, positive predictive value, and negative pre- dictive value of the nine scales under the two ELVO definitions were separately calculated. Statistical significance was defined as p < 0.05. We defined scales with high sensitivity as those with values higher than 80% and scales with low sensitivity as those with values lower than 70%. Others with sensitivity between 70- 80% were defined as scales with intermediate sensitivity.

1. Results
   1. Participants

During the study period, a total of 7197 patients arrived at the ED with a final diagnosis of acute stroke or TIA in the registry. Among them, 2646 patients with stroke arrived at the ED within 6 h of symp- tom onset. After excluding patients with hemorrhagic stroke and tran- sient ischemic attack (n = 1202), missing NIHSS scores (n = 99), and without brain imaging that meets our study criteria (n = 114), 1231 pa- tients were included in the final analysis. The study design is shown as a flowchart in Fig. 1. The average age of the patients was 70.5 years, and

Fig. 1. Flowchart of the study design. Abbreviations: ED, emergency department; TIA, transient ischemic attack; NIHSS, National Institutes of Health Stroke Scale.

57.2% of patients were male. Approximately half of the patients were brought to our hospital by EMS. The most common comorbidity was hy- pertension (70.1%), followed by hyperlipidemia (34.6%) and diabetes mellitus (33.2%). Median NIHSS score was seven. Approximately one- fourth of the patients received intravenous thrombolysis, while 11.6% underwent EVT. The characteristics of the patients are presented in Table 1.

• 1. Performance of scales

The prediction performance of the nine different prehospital stroke scales with the large vessel definition of ICA and M1 is shown in Table 2 and Supplementary Fig. S1. The AUROC of the nine scales mostly showed no significant differences, ranging from 74.74 to 80.65. The AUROC of the FAST-ED was significantly higher than that of the VAN screening tool. However, the sensitivities of the nine scales varied. The EMSA had the highest sensitivity (93.68%) but also showed the lowest specificity (38.37%) among the nine scales under the first ELVO defini- tion. In contrast, the 3-ISS had the lowest sensitivity (44.56%) but the highest specificity (88.9%). Under the large vessel definition of ICA and M1, four scales belonged to the high-sensitivity (>80%) scales, namely, G-FAST, VAN screening tool, PASS, and EMSA. Two scales, namely, 3-ISS and FPSS, belonged to the low-sensitivity (<70%) scales.

The prediction performance with the second large vessel definition is shown in Table 3 and Supplementary Fig. S2. Under the second large vessel definition, only the AUROCs of the FAST-ED and the FPSS were significantly higher than those of the VAN screening tool. There were no significant differences in the AUROC among other scales. Differ- ent sensitivities and specificities among the nine prehospital stroke scales were also noted. The G-FAST, VAN screening tool, and EMSA were the high-sensitivity (>80%) scales, while 3-ISS, and FPSS were the low-sensitivity (<70%) scales.

Table 1

Characteristics of study patients.

Characteristics	mean (SD)/median [q1, q3]
Age (years)	70.5 (13.64)
Male sex, n (%)	704 (57.2%)
Arrival by EMS Underlying disease, n (%)	584 (47.4%)
Old stroke	283 (23.0%)
Atrial fibrillation Congestive heart failure	359 (29.2%) 97 (7.9%)

Cardiovascular disease 224 (18.2%)

Carotid stenosis 46 (3.7%)

peripheral arterial occlusive disease 24 (1.9%)

Diabetes mellitus 409 (33.2%)

Hypertension 863 (70.1%)

Prosthetic valve 73 (5.9%)

Hyperlipidemia 426 (34.6%)

Smoking 213 (17.3%)

NIHSS score 7 [3, 17]

Intravenous thrombolysis 307 (24.9%)

Endovascular thrombectomy 143 (11.6%)

Abbreviations: SD, standard deviation; EMS, emergency medical service; NIHSS, National Institutes of Health Stroke Scale.

1. Discussion

This is the first study to externally validate different prehospital stroke scales to identify ELVO using the clinical data of patients in Taiwan. We compared nine different scales and found that they mostly had similar moderate to good discriminative ability of LVO under the two ELVO definitions. However, the sensitivities and specificities were quite different. The EMSA, G-FAST, and VAN Screening tools were iden- tified as high-sensitivity scales, while the 3-ISS and FPSS were low- sensitivity scales. A study reported that high-sensitivity scales are better

Table 2

Accuracy of different stroke scales with the large vessel definition of ICA and M1.

Stroke scales	Sensitivity	Specificity	Positive predictive value	Negative predictive value	Accuracy	AUROC (%) (95% confidence interval)
G-FAST	240/285 (84.21%)	563/946 (59.51%)	240/623 (38.52%)	563/608 (92.6%)	803/1231 (65.23%)	78.42 (75.64,81.20)
3-ISS	127/285 (44.56%)	841/946 (88.90%)	127/232 (54.74%)	841/999 (84.18%)	968/1231 (78.64%)	77.13 (74.07,80.18)
CPSSS	219/285 (76.84%)	680/946 (71.88%)	219/485 (45.15%)	680/746 (91.15%)	899/1231 (73.03%)	78.13 (75.17,81.09)
VAN	238/285 (83.51%)	624/946 (65.96%)	238/560 (42.5%)	624/671 (93%)	862/1231 (70.02%)	74.74 (72.1,77.37)
PASS	230/285 (80.7%)	652/946 (68.92%)	230/524 (43.89%)	652/707 (92.22%)	882/1231 (71.65%)	77.87 (74.98,80.75)
FAST-ED	216/285 (75.79%)	730/946 (77.17%)	216/432 (50%)	730/799 (91.36%)	946/1231 (76.85%)	80.65 (77.79,83.52)
RACE	214/285 (75.09%)	720/946 (76.11%)	214/440 (48.64%)	720/791 (91.02%)	934/1231 (75.87%)	79.67 (76.81,82.53)
EMSA	267/285 (93.68%)	363/946 (38.37%)	267/850 (31.41%)	363/381 (95.28%)	630/1231 (51.18%)	79.5 (76.62,82.37)
FPSS	192/285 (67.37%)	716/946 (75.69%)	192/422 (45.50%)	716/809 (88.5%)	908/1231 (73.76%)	79.08 (76.25,81.92)

Sensitivity = number of ELVO patients with positive result of the scale/number of ELVO patients. Specificity = number of patients without ELVO whose result of the scale were negative/ number of patients without ELVO. Positive predictive value = number of ELVO patients with positive result of the scale/number of patients with positive result of the scale. Negative pre- dictive value = number of patients without ELVO whose result of the scale were negative/number of patients with negative result of the scale. Accuracy = number of patients with correct scale result/total number of patients. Abbreviations: AUROC, area under the receiver operating characteristic curve; G-FAST, gaze-face-arm-speech-time test; 3-ISS, 3-Item Stroke Scale; CPSSS, Cincinnati Prehospital Stroke Severity Scale; VAN, Vision, Aphasia, Neglect screening tool; PASS, Prehospital Acute Stroke Severity Scale; FAST-ED, Field Assessment Stroke Triage for Emergency Destination; RACE, The Rapid Arterial Occlusion Evaluation scale; EMSA, Emergency Medical Stroke Assessment; FPSS, Finnish Prehospital Stroke Scale; ELVO, emergent large vessel occlusion.

Table 3

Accuracy of different stroke scales with the large vessel definition of ICA, M1, M2, A1, P1, and basilar artery.

Stroke scales	Sensitivity	Specificity	Positive predictive value	Negative predictive value	Accuracy	AUROC (%) (95% confidence interval)
G-FAST	342/419 (81.62%)	531/812 (65.39%)	342/623 (54.9%)	531/608 (87.34%)	873/1231 (70.92%)	80.25 (77.74,82.77)
3-ISS	179/419 (42.72%)	759/812 (93.47%)	179/232 (77.16%)	759/999 (75.98%)	938/1231 (76.2%)	79.98 (77.3,82.67)
CPSSS	309/419 (73.75%)	636/812 (78.33%)	309/485 (63.71%)	636/746 (85.25%)	945/1231 (76.77%)	80.58 (78.01,83.15)
VAN	338/419 (80.67%)	590/812 (72.66%)	338/560 (60.36%)	590/671 (87.93%)	928/1231 (75.39%)	76.66 (74.23,79.10)
PASS	321/419 (76.61%)	609/812 (75%)	321/524 (61.26%)	609/707 (86.14%)	930/1231 (75.55%)	79.78 (77.21,82.35)
FAST-ED	303/419 (72.32%)	683/812 (84.11%)	303/432 (70.14%)	683/799 (85.48%)	986/1231 (80.1%)	82.77 (80.26,85.27)
RACE	294/419 (70.17%)	666/812 (82.02%)	294/440 (66.82%)	666/791 (84.2%)	960/1231 (77.99%)	81.33 (78.78,83.87)
EMSA	384/419 (91.65%)	346/812 (42.61%)	384/850 (45.18%)	346/381 (90.81%)	730/1231 (59.3%)	79.41 (76.80,82.03)
FPSS	275/419 (65.63%)	665/812 (81.9%)	275/422 (65.17%)	665/809 (82.2%)	940/1231 (76.36%)	81.64 (79.13,84.14)

Sensitivity = number of ELVO patients with positive result of the scale/number of ELVO patients. Specificity = number of patients without ELVO whose result of the scale were negative/ number of patients without ELVO. Positive predictive value = number of ELVO patients with positive result of the scale/number of patients with positive result of the scale. Negative pre- dictive value = number of patients without ELVO whose result of the scale were negative/number of patients with negative result of the scale. Accuracy = number of patients with correct scale result/total number of patients. Abbreviations: AUROC, area under the receiver operating characteristic curve; G-FAST, gaze-face-arm-speech-time test; 3-ISS, 3-Item Stroke Scale; CPSSS, Cincinnati Prehospital Stroke Severity Scale; VAN, Vision, Aphasia, Neglect screening tool; PASS, Prehospital Acute Stroke Severity Scale; FAST-ED, Field Assessment Stroke Triage for Emergency Destination; RACE, The Rapid Arterial Occlusion Evaluation scale; EMSA, Emergency Medical Stroke Assessment; FPSS, Finnish Prehospital Stroke Scale; ELVO, emergent large vessel occlusion.

in the setting of shorter Transfer time to a comprehensive stroke center, while high-specificity scales are for the contrasting situation [34]. We recommend that stakeholders in the community should choose suitable ELVO scales according to their own system conditions.

Further, the scale for identifying ELVO should be easy enough for Prehospital personnel to remember and to complete evaluation as soon as possible when utilized. To achieve this goal, some ELVO scales have the same items as those in the tools used to identify patients with stroke. For example, EMSA and G-FAST have several items that are the same as those of the Cincinnati Prehospital Stroke Scale . Such characteristics make Prehospital providers complete their assessment for ELVO identification faster for patients suspected of suffering from stroke using CPSS. It also makes it easy to be remem- bered. Some ELVO scales adopt dichotomized items; therefore, the ex- aminer does not need to distinguish between mild-to-moderate and severe deficits [29], and inter-rater reliability might be improved. This can explain why CPSSS and PASS are easier to perform and interpret than 3-ISS because the items in the former are dichotomized. In addi- tion, some Neurological deficits such as visual quadrant field cut or ne- glect are subtle symptoms, taking more time for the examiner to learn how to recognize these symptoms and to feel confident about performing the examinations. Therefore, the VAN screening tool, which included these items, had less chance to be accepted as a tool in a prehospital setting [18].

In our study, the sensitivities of EMSA, G-FAST, and VAN screening

tool were relatively high. Scheitz et al. (2017) demonstrated G-FAST

as a high-sensitivity scale, with a sensitivity of 88.7% by the ELVO defini- tion of ICA and M1 [34]. In other studies that included patients with stroke and stroke mimic, the G-FAST and EMSA also had good sensitivity scores [29,34]. In our study, the 3-ISS scale demonstrated the highest specificity but a low sensitivity. Previous studies have reported similar prediction values in patients who were verified to have IS or suspected of having stroke [29,31,33,34]. Although the sensitivity of the VAN screening tool was shown to be 100% in the original study [18], the sen- sitivity was much lower in our study (83.5%). The original FPSS study showed a relatively low sensitivity of 54.3% [33], which was similar to that in our study.

1. Limitation

There are some limitations in our study. First, the patients of our study were recruited from the in-hospital cohort with a final diagnosis of acute IS. The ability of these scales to distinguish acute IS from Stroke mimics could not be determined from our study. Second, we compared the prehospital scales based on the NIHSS score. The scale containing additional items other than the NIHSS score, such as Los Angeles Motor Scale which needs grip strength, could not be evaluated in our study [16]. Third, the NIHSS score in our study was evaluated by a resi- dent neurologist or a board-certificated neurologist rather than prehospital providers. Further interventional validation studies in prehospital settings are warranted.

1. Conclusion

The nine scales mostly had similar moderate to good discriminative ability to identify ELVO but had quite different sensitivities and specific- ities. Stakeholders in communities should choose suitable ELVO scales according to their own system environment.

Credit author statement Hsieh MJ, Tang SC, Lee CW: Conceptualization; Lin YH, Tang SC, Tsai

LK, Lee CW: Data curation; Hsieh MJ: Formal analysis; Tang SC, Hsieh

MJ: Funding acquisition; Chiu YC: Investigation; Chiu YC, Hsieh MJ, Jeng JS: Methodology; Lin YH, Jeng JS: Project administration; Hsieh MJ, Lin YH, Tang SC: Resources; Chiang WC, Lee YC: Software; Sun JT, Tsai LK, Jeng JS: Supervision; Sun JT, Jeng JS: Validation; Hsieh MJ, Lee YC: Visualization; Chiu YC, Hsieh MJ: Roles/Writing - original draft; Tang SC, Tsai LK, Jeng JS: Writing - review & editing.

Declaration of Competing Interest

All authors disclose that there were no potential interests.

Acknowledgments

We acknowledge the statistical assistance provided by the Taiwan Clinical Trial Bioinformatics and Statistical Center, Training Center, and Pharmacogenomics Laboratory, and the Department of Medical Re- search of National Taiwan University Hospital. The work was supported by the Taiwan Ministry of Science and Technology (MOST 108-2314-B- 002-131) and National Taiwan University Hospital (108-09).

Appendix A. Supplementary data

Supplementary data to this article can be found online at https://doi. org/10.1016/j.ajem.2020.12.011.

References

1. Berkhemer OA, Fransen PS, Beumer D, et al. A randomized trial of intraarterial treat- ment for acute ischemic stroke. N Engl J Med. 2015;372:11-20. https://doi.org/10. 1056/NEJMoa1411587.
2. Campbell BC, Mitchell PJ, Kleinig TJ, et al. Endovascular therapy for ischemic stroke with perfusion-imaging selection. N Engl J Med. 2015;372:1009-18. https://doi. org/10.1056/NEJMoa1414792.
3. Goyal M, Demchuk AM, Menon BK, et al. Randomized assessment of rapid endovascular treatment of ischemic stroke. N Engl J Med. 2015;372:1019-30. https://doi.org/10.1056/NEJMoa1414905.
4. Saver JL, Goyal M, Bonafe A, et al. Stent-retriever thrombectomy after intravenous t- PA vs. t-PA alone in stroke. N Engl J Med. 2015;372:2285-95. https://doi.org/10. 1056/NEJMoa1415061.
5. Jovin TG, Chamorro A, Cobo E, et al. Thrombectomy within 8 hours after symptom onset in ischemic stroke. N Engl J Med. 2015;372:2296-306. https://doi.org/10. 1056/NEJMoa1503780.
6. Sheth SA, Jahan R, Gralla J, et al. Time to endovascular reperfusion and degree of dis- ability in acute stroke. Ann Neurol. 2015;78:584-93. https://doi.org/10.1002/ana. 24474.
7. Menon BK, Sajobi TT, Zhang Y, et al. Analysis of workflow and time to treatment on thrombectomy outcome in the ESCAPE randomized controlled trial. Circulation. 2016;133:2279-86. https://doi.org/10.1161/CIRCULATIONAHA.115.019983.
8. Goyal M, Jadhav AP, Bonafe A, et al. Analysis of workflow and time to treatment and the effects on outcome in endovascular treatment of acute ischemic stroke: results from the SWIFT PRIME randomized controlled trial. Radiology. 2016;279:888-97. https://doi.org/10.1148/radiol.2016160204.
9. Fransen PS, Berkhemer OA, Lingsma HF, et al. Time to reperfusion and treatment ef- fect for acute ischemic stroke: a randomized clinical trial. JAMA Neurol. 2016;73: 190-6. https://doi.org/10.1001/jamaneurol.2015.3886.
10. Ng FC, Low E, Andrew E, et al. Deconstruction of interhospital transfer workflow in large vessel occlusion: real-world data in the thrombectomy era. Stroke. 2017;48: 1976-9. https://doi.org/10.1161/STROKEAHA.117.017235.
11. Shah S, Xian Y, Sheng S, et al. Use, temporal trends, and outcomes of endovascular therapy after interhospital transfer in the United States. Circulation. 2019;139: 1568-77. https://doi.org/10.1161/CIRCULATIONAHA.118.036509.
12. Pride GL, Fraser JF, Gupta R, et al. Prehospital care delivery and triage of stroke with emergent large vessel occlusion (ELVO): report of the Standards and Guidelines Committee of the Society of Neurointerventional Surgery. J Neurointerv Surg. 2017;9:802-12. https://doi.org/10.1136/neurintsurg-2016-012699.
13. Powers WJ, Rabinstein AA, Ackerson T, et al. Guidelines for the early management of patients with acute ischemic stroke: 2019 update to the 2018 guidelines for the early management of acute ischemic stroke: a guideline for healthcare professionals from the American Heart Association/American Stroke Association. Stroke. 2019;50: e344-418. https://doi.org/10.1161/STR.0000000000000211.
14. Heldner MR, Zubler C, Mattle HP, et al. National Institutes of Health Stroke Scale score and vessel occlusion in 2152 patients with acute ischemic stroke. Stroke. 2013;44:1153-7. https://doi.org/10.1161/STROKEAHA.111.000604.
15. Singer OC, Dvorak F, du Mesnil de Rochemont R, et al. A simple 3-item stroke scale: comparison with the National Institutes of Health stroke scale and prediction of middle cerebral artery occlusion. Stroke. 2005;36:773-6. https://doi.org/10.1161/ 01.STR.0000157591.61322.df.
16. Nazliel B, Starkman S, Liebeskind DS, et al. A brief prehospital stroke severity scale identifies ischemic stroke patients harboring persisting large arterial occlusions. Stroke. 2008;39:2264-7. https://doi.org/10.1161/STROKEAHA.107.508127.
17. Katz BS, McMullan JT, Sucharew H, et al. Design and validation of a prehospital scale to predict stroke severity: Cincinnati Prehospital Stroke Severity Scale. Stroke. 2015; 46:1508-12. https://doi.org/10.1161/STROKEAHA.115.008804.
18. Teleb MS, Ver Hage A, Carter J, et al. Stroke vision, aphasia, neglect (VAN) assessment-a novel emergent large vessel occlusion screening tool: pilot study and comparison with current clinical severity indices. J Neurointerv Surg. 2017;9: 122-6. https://doi.org/10.1136/neurintsurg-2015-012131.
19. Smith EE, Kent DM, Bulsara KR, et al. Accuracy of prediction instruments for diagnos- ing large vessel occlusion in individuals with suspected stroke: a systematic review for the 2018 guidelines for the early management of patients with acute ischemic stroke. Stroke. 2018;49:e111-22. https://doi.org/10.1161/STR.0000000000000160.
20. Jayaraman MV, Iqbal A, Silver B, et al. Developing a statewide protocol to ensure pa- tients with suspected emergent large vessel occlusion are directly triaged in the field to a comprehensive stroke center: how we did it. J Neurointerv Surg. 2017;9:330-2. https://doi.org/10.1136/neurintsurg-2016-012275.
21. Zaidi SF, Shawver J, Espinosa Morales A, et al. Stroke care: initial data from a county- based bypass protocol for patients with acute stroke. J Neurointerv Surg. 2017;9: 631-5. https://doi.org/10.1136/neurintsurg-2016-012476.
22. Hu HH, Sheng WY, Chu FL, et al. Incidence of stroke in Taiwan. Stroke. 1992;23: 1237-41. https://doi.org/10.1161/01.str.23.9.1237.
23. Hsieh FI, Chiou HY. Stroke: morbidity, risk factors, and care in Taiwan. J Stroke. 2014;

16:59-64. https://doi.org/10.5853/jos.2014.16.2.59.

1. Yip PK, Jeng JS, Lee TK, Chang YC, Huang ZS, Ng SK, et al. Subtypes of ischemic stroke: a hospital-based stroke registry in Taiwan (SCAN-IV). Stroke. 1997;28:2507-12. https://doi.org/10.1161/01.str.28.12.2507.
2. Lee HY, Hwang JS, Jeng JS, et al. Quality-adjusted Life expectancy (QALE) and loss of QALE for patients with ischemic stroke and intracerebral hemorrhage: a 13-year follow-up. Stroke. 2010;41:739-44. https://doi.org/10.1161/STROKEAHA.109. 573543.
3. The World Health Organization MONICA Project (monitoring trends and determi- nants in cardiovascular disease): a major international collaboration. WHO MONICA project principal investigators. J Clin Epidemiol. 1988;41:105-14. https:// doi.org/10.1016/0895-4356(88)90084-4.
4. Hsieh FI, Lien LM, Chen ST, et al. Get with the guidelines-stroke performance indica- tors: surveillance of stroke care in the Taiwan stroke registry: get with the guidelines-stroke in Taiwan. Circulation. 2010;122:1116-23. https://doi.org/10. 1161/CIRCULATIONAHA.110.936526.
5. Chu HJ, Tang SC, Lee CW, et al. Endovascular thrombectomy for acute ischemic stroke: a single-center experience in Taiwan. J Formos Med Assoc. 2018;117: 806-13. https://doi.org/10.1016/j.jfma.2017.09.016.
6. Gropen TI, Boehme A, Martin-Schild S, et al. Derivation and validation of the emer- gency medical stroke assessment and comparison of large vessel occlusion scales. J Stroke Cerebrovasc Dis. 2018;27:806-15. https://doi.org/10.1016/j. jstrokecerebrovasdis.2017.10.018.
7. Lima FO, Silva GS, Furie KL, Frankel MR, Lev MH, Camargo EC, et al. Field assessment stroke triage for emergency destination: a simple and accurate prehospital scale to detect large vessel occlusion strokes. Stroke. 2016;47:1997-2002. https://doi.org/ 10.1161/STROKEAHA.116.013301.
8. Hastrup S, Damgaard D, Johnsen SP, et al. Prehospital acute stroke severity scale to predict Large artery occlusion: design and comparison with other scales. Stroke. 2016;47:1772-6. https://doi.org/10.1161/STROKEAHA.115.012482.
9. Perez de la Ossa N, Carrera D, Gorchs M, et al. Design and validation of a prehospital stroke scale to predict large arterial occlusion: the rapid arterial occlusion evaluation scale. Stroke. 2014;45:87-91. https://doi.org/10.1161/STROKEAHA.113.003071.
10. Ollikainen J, Janhunen H, Tynkkynen J, et al. The finnish prehospital stroke scale de- tects thrombectomy and thrombolysis candidates-a propensity score-matched study. J Stroke Cerebrovasc Dis. 2018;27:771-7. https://doi.org/10.1016/j. jstrokecerebrovasdis.2017.10.015.
11. Scheitz JF, Abdul-Rahim AH, MacIsaac RL, et al. Clinical selection strategies to identify ischemic stroke patients with large anterior vessel occlusion: results from SITS-ISTR (Safe Implementation of Thrombolysis in Stroke International stroke thrombolysis Registry). Stroke. 2017;48:290-7. https://doi.org/10.1161/STROKEAHA.116.014431.
12. Collins GS, Ogundimu EO, Altman DG. Sample size considerations for the external validation of a multivariable prognostic model: a resampling study. Stat Med. 2016;35:214-26. https://doi.org/10.1002/sim.6787.